Time_weight in continuous aggregate

TimescaleDB and PostgreSQL
1

I am collecting power data from a set of sensors at uneven intervals. I want to create energy aggregations from the power data. The power data is stored in the following table:

CREATE TABLE power_rows (
    id INT GENERATED ALWAYS AS IDENTITY,
    sensor_id TEXT,
    "timestamp" TIMESTAMPTZ,
    "value" INT NOT NULL,
    PRIMARY KEY (sensor_id, "timestamp")
) WITH (
    timescaledb.hypertable = true,
    timescaledb.partition_column = 'timestamp',
    timescaledb.segmentby = 'sensor_id'
);

Now, I want run the following query to gather 1 minute aggregations:

    SELECT
        sensor_id,
        time AS from,
        time + INTERVAL '1 minute' AS to,
        interpolated_integral(
            tws,
            time,
            '1 minute',
            LAG(tws) OVER (PARTITION BY sensor_id ORDER BY time),
            LEAD(tws) OVER (PARTITION BY sensor_id ORDER BY time),
            'hours'
        ) / 1000 AS value
    FROM (
        SELECT
            sensor_id,
            time_bucket('1 minute', timestamp) AS time,
            time_weight('Linear', timestamp, value) AS tws
        FROM power_rows
        GROUP BY sensor_id, time
    ) t;

which works as intended. However when wrapping the query above in a materialized view

CREATE MATERIALIZED VIEW energy_rows_1min
WITH (
    timescaledb.continuous = true,
    timescaledb.materialized_only = false
) AS
   <query from above>
    GROUP BY sensor_id, "from";

I get the following error
ERROR: column "t.tws" must appear in the GROUP BY clause or be used in an aggregate function

How can make the query be a continuous aggregate?
(I have set enabled experimental window function support by setting timescaledb.enable_cagg_window_functions)